much as you are appreciative of the deflators in your life (whose ranks i count myself among) i am appreciative of posts like this that let me believe something incredible *is* happening. thank you frank!
I might be misreading, but it kind of sounds like the problem, such as it is, is not that the deflators are wrong, but that the arguments they make sometimes have the side effect of downplaying how cool and amazing this all is.
I think it's possible to be a deflator who still finds these breakthroughs meaningful and significant (in fact, that describes me pretty well), though. Just meaningful and significant along some other axis than doom/AGI.
The meaningful and significant thing, to me, is that we've actually managed to inject a teeny tiny bit of structure into the giant frothing pot that is Big Data. And the reason I find this cool rather than concerning is that this last crucial bit, this spark, is still coming from us.
I first understood "the ethnomethodological flip" as an insight into the workings of my own intelligence, but lately I see it as evidence in favor of your position here (which I certainly share). I often think of that book chapter about learning board game rules, which perhaps paints a picture of how a predictive mess can come to approximate/emulate symbolic logic.
Are you going to brand your position a la "unpluggers"/"deflators"? Is that a spoiler for the post series?
I'd argue there WAS one big , huge, insight that marked the absolute inflection point in these things. Attention. If you look at the history of these things, they just didnt start getting good until that "Attention is all you need" paper that gave us transformers, then everything clicked into place and bam, I can now argue with a robot about Kirk vs Picard (the answer, by the way, is Picard).
What bakes my noodle, is what Heidegger said about consciousness, that consciousness is always being conscious OF something. Ie, its just "paying attention". No, I dont happen to think these things are 'conscious' in any way phenomenologically recognizable to our experience of it, but SOMETHING is happening here. I feel we are on the brink.
Frank, thanks for the shout out! In a way, your post reminds me of what journalists had to say about personal computers in the early 1980's. I think you're chipping away at the veneer, to see if there's something there.
Although the engineering of LLM is beyond me, I can't help but shake my head when I read about how even the architects can't easily figure out how their golem reached a given conclusion.
Do you think that tiny howl has the potential to signify something magnificent? Teflon was an accident, too. Then it became a thing, until finally, we learned that it's actually bad for us and the environment!
Unpluggers etc, Part 2: Deflategate
much as you are appreciative of the deflators in your life (whose ranks i count myself among) i am appreciative of posts like this that let me believe something incredible *is* happening. thank you frank!
I might be misreading, but it kind of sounds like the problem, such as it is, is not that the deflators are wrong, but that the arguments they make sometimes have the side effect of downplaying how cool and amazing this all is.
I think it's possible to be a deflator who still finds these breakthroughs meaningful and significant (in fact, that describes me pretty well), though. Just meaningful and significant along some other axis than doom/AGI.
The meaningful and significant thing, to me, is that we've actually managed to inject a teeny tiny bit of structure into the giant frothing pot that is Big Data. And the reason I find this cool rather than concerning is that this last crucial bit, this spark, is still coming from us.
my god, man. you're a poet. keep writing, please. absolutely wonderful stuff. i can relate.
I first understood "the ethnomethodological flip" as an insight into the workings of my own intelligence, but lately I see it as evidence in favor of your position here (which I certainly share). I often think of that book chapter about learning board game rules, which perhaps paints a picture of how a predictive mess can come to approximate/emulate symbolic logic.
Are you going to brand your position a la "unpluggers"/"deflators"? Is that a spoiler for the post series?
I'd argue there WAS one big , huge, insight that marked the absolute inflection point in these things. Attention. If you look at the history of these things, they just didnt start getting good until that "Attention is all you need" paper that gave us transformers, then everything clicked into place and bam, I can now argue with a robot about Kirk vs Picard (the answer, by the way, is Picard).
What bakes my noodle, is what Heidegger said about consciousness, that consciousness is always being conscious OF something. Ie, its just "paying attention". No, I dont happen to think these things are 'conscious' in any way phenomenologically recognizable to our experience of it, but SOMETHING is happening here. I feel we are on the brink.
Frank, thanks for the shout out! In a way, your post reminds me of what journalists had to say about personal computers in the early 1980's. I think you're chipping away at the veneer, to see if there's something there.
Although the engineering of LLM is beyond me, I can't help but shake my head when I read about how even the architects can't easily figure out how their golem reached a given conclusion.
Do you think that tiny howl has the potential to signify something magnificent? Teflon was an accident, too. Then it became a thing, until finally, we learned that it's actually bad for us and the environment!